{{MODEL_NAME}}

{{MODEL_FILE}} · HW target: {{HW_TARGET}} · Verification mode: {{VERIFICATION_MODE}} · Goal: {{GOAL}} · Run completed {{RUN_DATE}}

Headline Outcomes

{{HEADLINE_CARDS}}

Iterations (Model-Change → Re-measurement Cycles)

{{ITERATION_ROWS}}
Iter Version Changes Applied {{METRIC1_LABEL}} {{METRIC2_LABEL}} Notes

{{ITERATION_FOOTNOTE}}

Evolution Tree

Each node is a git snapshot in the workspace repo at {{GIT_REPO_PATH}}. Sibling branches off a parent represent either (a) candidates that were auto-rejected by the Goal-Axis Acceptance Gate (Gate Rejected — dashed red connector) or (b) versions the user asked to roll back from after re-measurement (User-Requested Revert — solid amber connector with a clickable back-pointer to the restored target). Both kinds of leaf remain queryable via their git tag but are not part of the live accepted lineage. {{TREE_PREAMBLE_NOTE}}

Original (v0) Baseline (config normalisation) Accepted (gate passed) Gate Rejected (auto-revert) User-Requested Revert
{{EVOLUTION_TREE_NODES}}

Tree rendering conventions: Each node is positioned by its parentVersion, so a parent with multiple attempts shows its children side-by-side as sibling branches. Nodes with revertCause are always leaves — the tree never descends past them. A dashed red connector labelled "GATE REJECTED" marks an auto-revert by the Goal-Axis Acceptance Gate (the offending axis delta is highlighted in red inside the leaf). A solid amber connector labelled "USER REVERT" marks a user-requested rollback after re-measurement; each such leaf also carries a clickable ↶ back-pointer in its details body that jumps to the restored target version (the live workspace is now at that target, and any further optimization branches off it, not off the revert leaf). Both kinds of leaf retain a git tag so the abandoned state can be re-loaded for inspection via eco_revert.

Token Usage (end-to-end workflow)

{{TOKEN_CARDS}}

Source: .eco_diagnostics/eco_optimization_report.md

Sim-vs-SIL Correctness Verification

Verdict
{{VERIFY_VERDICT}}
{{VERIFY_VERDICT_SUB}}
Max abs error
{{VERIFY_MAX_ERR}}
{{VERIFY_MAX_ERR_SUB}}

{{VERIFY_NARRATIVE}}

Artifacts: {{VERIFY_ARTIFACTS}}

Further Recommendations

These are additional optimizations identified during the run that were not applied — either because they require manual effort, domain-specific decisions, or fall outside the automated workflow scope.

{{PREFERENCES_SECTION}}